Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / 3669299acdb4239d1c515e1b7a16480c50473cdc / . / docs / Stacker.html
blob: fca4b0bc64a62b985f1b691529fb9db4cd523904 [file] [log] [blame]
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]3<html>
4<head>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]5 <title>Stacker: An Example Of Using LLVM</title>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]6 <link rel="stylesheet" href="llvm.css" type="text/css">
7</head>
8<body>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]9
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]10<div class="doc_title">Stacker: An Example Of Using LLVM</div>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]11
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]12<ol>
13 <li><a href="#abstract">Abstract</a></li>
14 <li><a href="#introduction">Introduction</a></li>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]15 <li><a href="#lessons">Lessons I Learned About LLVM</a>
16 <ol>
17 <li><a href="#value">Everything's a Value!</a></li>
18 <li><a href="#terminate">Terminate Those Blocks!</a></li>
19 <li><a href="#blocks">Concrete Blocks</a></li>
20 <li><a href="#push_back">push_back Is Your Friend</a></li>
21 <li><a href="#gep">The Wily GetElementPtrInst</a></li>
22 <li><a href="#linkage">Getting Linkage Types Right</a></li>
23 <li><a href="#constants">Constants Are Easier Than That!</a></li>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]24 </ol></li>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]25 <li><a href="#lexicon">The Stacker Lexicon</a>
26 <ol>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]27 <li><a href="#stack">The Stack</a></li>
28 <li><a href="#punctuation">Punctuation</a></li>
29 <li><a href="#comments">Comments</a></li>
30 <li><a href="#literals">Literals</a></li>
31 <li><a href="#words">Words</a></li>
32 <li><a href="style">Standard Style</a></li>
33 <li><a href="#builtins">Built-Ins</a></li>
34 </ol></li>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]35 <li><a href="#example">Prime: A Complete Example</a></li>
36 <li><a href="#internal">Internal Code Details</a>
37 <ol>
38 <li><a href="#directory">The Directory Structure </a></li>
39 <li><a href="#lexer">The Lexer</a></li>
40 <li><a href="#parser">The Parser</a></li>
41 <li><a href="#compiler">The Compiler</a></li>
42 <li><a href="#runtime">The Runtime</a></li>
43 <li><a href="#driver">Compiler Driver</a></li>
44 <li><a href="#tests">Test Programs</a></li>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]45 <li><a href="#exercise">Exercise</a></li>
46 <li><a href="#todo">Things Remaining To Be Done</a></li>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]47 </ol></li>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]48</ol>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]49
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]50<div class="doc_text">
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]51<p><b>Written by <a href="mailto:rspencer@x10sys.com">Reid Spencer</a></b></p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]52</div>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]53
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]54<!-- ======================================================================= -->
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]55<div class="doc_section"><a name="abstract">Abstract</a></div>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]56<div class="doc_text">
57<p>This document is another way to learn about LLVM. Unlike the
58<a href="LangRef.html">LLVM Reference Manual</a> or
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]59<a href="ProgrammersManual.html">LLVM Programmer's Manual</a>, here we learn
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]60about LLVM through the experience of creating a simple programming language
61named Stacker. Stacker was invented specifically as a demonstration of
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]62LLVM. The emphasis in this document is not on describing the
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]63intricacies of LLVM itself but on how to use it to build your own
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]64compiler system.</p>
65</div>
66<!-- ======================================================================= -->
67<div class="doc_section"> <a name="introduction">Introduction</a> </div>
68<div class="doc_text">
69<p>Amongst other things, LLVM is a platform for compiler writers.
70Because of its exceptionally clean and small IR (intermediate
71representation), compiler writing with LLVM is much easier than with
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]72other system. As proof, I wrote the entire compiler (language definition,
73lexer, parser, code generator, etc.) in about <em>four days</em>!
74That's important to know because it shows how quickly you can get a new
75language running when using LLVM. Furthermore, this was the <em >first</em>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]76language the author ever created using LLVM. The learning curve is
77included in that four days.</p>
78<p>The language described here, Stacker, is Forth-like. Programs
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]79are simple collections of word definitions, and the only thing definitions
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]80can do is manipulate a stack or generate I/O. Stacker is not a "real"
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]81programming language; it's very simple. Although it is computationally
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]82complete, you wouldn't use it for your next big project. However,
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]83the fact that it is complete, it's simple, and it <em>doesn't</em> have
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]84a C-like syntax make it useful for demonstration purposes. It shows
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]85that LLVM could be applied to a wide variety of languages.</p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]86<p>The basic notions behind stacker is very simple. There's a stack of
87integers (or character pointers) that the program manipulates. Pretty
88much the only thing the program can do is manipulate the stack and do
89some limited I/O operations. The language provides you with several
90built-in words that manipulate the stack in interesting ways. To get
91your feet wet, here's how you write the traditional "Hello, World"
92program in Stacker:</p>
93<p><code>: hello_world "Hello, World!" &gt;s DROP CR ;<br>
94: MAIN hello_world ;<br></code></p>
95<p>This has two "definitions" (Stacker manipulates words, not
96functions and words have definitions): <code>MAIN</code> and <code>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]97hello_world</code>. The <code>MAIN</code> definition is standard; it
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]98tells Stacker where to start. Here, <code>MAIN</code> is defined to
99simply invoke the word <code>hello_world</code>. The
100<code>hello_world</code> definition tells stacker to push the
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]101<code>"Hello, World!"</code> string on to the stack, print it out
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]102(<code>&gt;s</code>), pop it off the stack (<code>DROP</code>), and
103finally print a carriage return (<code>CR</code>). Although
104<code>hello_world</code> uses the stack, its net effect is null. Well
105written Stacker definitions have that characteristic. </p>
106<p>Exercise for the reader: how could you make this a one line program?</p>
107</div>
108<!-- ======================================================================= -->
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]109<div class="doc_section"><a name="lessons"></a>Lessons I Learned About LLVM</div>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]110<div class="doc_text">
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]111<p>Stacker was written for two purposes: </p>
112<ol>
113 <li>to get the author over the learning curve, and</li>
114 <li>to provide a simple example of how to write a compiler using LLVM.</li>
115</ol>
116<p>During the development of Stacker, many lessons about LLVM were
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]117learned. Those lessons are described in the following subsections.<p>
118</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]119<!-- ======================================================================= -->
120<div class="doc_subsection"><a name="value"></a>Everything's a Value!</div>
121<div class="doc_text">
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]122<p>Although I knew that LLVM uses a Single Static Assignment (SSA) format,
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]123it wasn't obvious to me how prevalent this idea was in LLVM until I really
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]124started using it. Reading the <a href="ProgrammersManual.html">
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]125Programmer's Manual</a> and <a href="LangRef.html">Language Reference</a>,
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]126I noted that most of the important LLVM IR (Intermediate Representation) C++
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]127classes were derived from the Value class. The full power of that simple
128design only became fully understood once I started constructing executable
129expressions for Stacker.</p>
130<p>This really makes your programming go faster. Think about compiling code
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]131for the following C/C++ expression: <code>(a|b)*((x+1)/(y+1))</code>. Assuming
132the values are on the stack in the order a, b, x, y, this could be
133expressed in stacker as: <code>1 + SWAP 1 + / ROT2 OR *</code>.
134You could write a function using LLVM that computes this expression like this: </p>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]135<pre><code>
136Value*
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]137expression(BasicBlock* bb, Value* a, Value* b, Value* x, Value* y )
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]138{
139 Instruction* tail = bb->getTerminator();
140 ConstantSInt* one = ConstantSInt::get( Type::IntTy, 1);
141 BinaryOperator* or1 =
Chris Lattner0b404c82003-11-25 01:44:27 +0000[diff] [blame]142 BinaryOperator::create( Instruction::Or, a, b, "", tail );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]143 BinaryOperator* add1 =
Chris Lattner0b404c82003-11-25 01:44:27 +0000[diff] [blame]144 BinaryOperator::create( Instruction::Add, x, one, "", tail );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]145 BinaryOperator* add2 =
Chris Lattner0b404c82003-11-25 01:44:27 +0000[diff] [blame]146 BinaryOperator::create( Instruction::Add, y, one, "", tail );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]147 BinaryOperator* div1 =
Chris Lattner0b404c82003-11-25 01:44:27 +0000[diff] [blame]148 BinaryOperator::create( Instruction::Div, add1, add2, "", tail);
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]149 BinaryOperator* mult1 =
Chris Lattner0b404c82003-11-25 01:44:27 +0000[diff] [blame]150 BinaryOperator::create( Instruction::Mul, or1, div1, "", tail );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]151
152 return mult1;
153}
154</code></pre>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]155<p>"Okay, big deal," you say? It is a big deal. Here's why. Note that I didn't
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]156have to tell this function which kinds of Values are being passed in. They could be
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]157<code>Instruction</code>s, <code>Constant</code>s, <code>GlobalVariable</code>s, or
158any of the other subclasses of <code>Value</code> that LLVM supports.
159Furthermore, if you specify Values that are incorrect for this sequence of
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]160operations, LLVM will either notice right away (at compilation time) or the LLVM
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]161Verifier will pick up the inconsistency when the compiler runs. In either case
162LLVM prevents you from making a type error that gets passed through to the
163generated program. This <em>really</em> helps you write a compiler that
164always generates correct code!<p>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]165<p>The second point is that we don't have to worry about branching, registers,
166stack variables, saving partial results, etc. The instructions we create
167<em>are</em> the values we use. Note that all that was created in the above
168code is a Constant value and five operators. Each of the instructions <em>is</em>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]169the resulting value of that instruction. This saves a lot of time.</p>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]170<p>The lesson is this: <em>SSA form is very powerful: there is no difference
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]171between a value and the instruction that created it.</em> This is fully
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]172enforced by the LLVM IR. Use it to your best advantage.</p>
173</div>
174<!-- ======================================================================= -->
175<div class="doc_subsection"><a name="terminate"></a>Terminate Those Blocks!</div>
176<div class="doc_text">
177<p>I had to learn about terminating blocks the hard way: using the debugger
178to figure out what the LLVM verifier was trying to tell me and begging for
179help on the LLVMdev mailing list. I hope you avoid this experience.</p>
180<p>Emblazon this rule in your mind:</p>
181<ul>
182 <li><em>All</em> <code>BasicBlock</code>s in your compiler <b>must</b> be
183 terminated with a terminating instruction (branch, return, etc.).
184 </li>
185</ul>
186<p>Terminating instructions are a semantic requirement of the LLVM IR. There
187is no facility for implicitly chaining together blocks placed into a function
188in the order they occur. Indeed, in the general case, blocks will not be
189added to the function in the order of execution because of the recursive
190way compilers are written.</p>
191<p>Furthermore, if you don't terminate your blocks, your compiler code will
192compile just fine. You won't find out about the problem until you're running
193the compiler and the module you just created fails on the LLVM Verifier.</p>
194</div>
195<!-- ======================================================================= -->
196<div class="doc_subsection"><a name="blocks"></a>Concrete Blocks</div>
197<div class="doc_text">
198<p>After a little initial fumbling around, I quickly caught on to how blocks
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]199should be constructed. In general, here's what I learned:
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]200<ol>
201 <li><em>Create your blocks early.</em> While writing your compiler, you
202 will encounter several situations where you know apriori that you will
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]203 need several blocks. For example, if-then-else, switch, while, and for
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]204 statements in C/C++ all need multiple blocks for expression in LVVM.
205 The rule is, create them early.</li>
206 <li><em>Terminate your blocks early.</em> This just reduces the chances
207 that you forget to terminate your blocks which is required (go
208 <a href="#terminate">here</a> for more).
209 <li><em>Use getTerminator() for instruction insertion.</em> I noticed early on
210 that many of the constructors for the Instruction classes take an optional
211 <code>insert_before</code> argument. At first, I thought this was a mistake
212 because clearly the normal mode of inserting instructions would be one at
213 a time <em>after</em> some other instruction, not <em>before</em>. However,
214 if you hold on to your terminating instruction (or use the handy dandy
215 <code>getTerminator()</code> method on a <code>BasicBlock</code>), it can
216 always be used as the <code>insert_before</code> argument to your instruction
217 constructors. This causes the instruction to automatically be inserted in
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]218 the RightPlace&trade; place, just before the terminating instruction. The
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]219 nice thing about this design is that you can pass blocks around and insert
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]220 new instructions into them without ever knowing what instructions came
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]221 before. This makes for some very clean compiler design.</li>
222</ol>
223<p>The foregoing is such an important principal, its worth making an idiom:</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]224<pre>
225BasicBlock* bb = new BasicBlock();
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]226bb->getInstList().push_back( new Branch( ... ) );
227new Instruction(..., bb->getTerminator() );
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]228</pre>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]229<p>To make this clear, consider the typical if-then-else statement
230(see StackerCompiler::handle_if() method). We can set this up
231in a single function using LLVM in the following way: </p>
232<pre>
233using namespace llvm;
234BasicBlock*
235MyCompiler::handle_if( BasicBlock* bb, SetCondInst* condition )
236{
237 // Create the blocks to contain code in the structure of if/then/else
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]238 BasicBlock* then_bb = new BasicBlock();
239 BasicBlock* else_bb = new BasicBlock();
240 BasicBlock* exit_bb = new BasicBlock();
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]241
242 // Insert the branch instruction for the "if"
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]243 bb->getInstList().push_back( new BranchInst( then_bb, else_bb, condition ) );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]244
245 // Set up the terminating instructions
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]246 then->getInstList().push_back( new BranchInst( exit_bb ) );
247 else->getInstList().push_back( new BranchInst( exit_bb ) );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]248
249 // Fill in the then part .. details excised for brevity
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]250 this->fill_in( then_bb );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]251
252 // Fill in the else part .. details excised for brevity
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]253 this->fill_in( else_bb );
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]254
255 // Return a block to the caller that can be filled in with the code
256 // that follows the if/then/else construct.
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]257 return exit_bb;
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]258}
259</pre>
260<p>Presumably in the foregoing, the calls to the "fill_in" method would add
261the instructions for the "then" and "else" parts. They would use the third part
262of the idiom almost exclusively (inserting new instructions before the
263terminator). Furthermore, they could even recurse back to <code>handle_if</code>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]264should they encounter another if/then/else statement, and it will just work.</p>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]265<p>Note how cleanly this all works out. In particular, the push_back methods on
266the <code>BasicBlock</code>'s instruction list. These are lists of type
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]267<code>Instruction</code> (which is also of type <code>Value</code>). To create
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]268the "if" branch we merely instantiate a <code>BranchInst</code> that takes as
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]269arguments the blocks to branch to and the condition to branch on. The
270<code>BasicBlock</code> objects act like branch labels! This new
271<code>BranchInst</code> terminates the <code>BasicBlock</code> provided
272as an argument. To give the caller a way to keep inserting after calling
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]273<code>handle_if</code>, we create an <code>exit_bb</code> block which is
274returned
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]275to the caller. Note that the <code>exit_bb</code> block is used as the
276terminator for both the <code>then_bb</code> and the <code>else_bb</code>
277blocks. This guarantees that no matter what else <code>handle_if</code>
278or <code>fill_in</code> does, they end up at the <code>exit_bb</code> block.
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]279</p>
280</div>
281<!-- ======================================================================= -->
282<div class="doc_subsection"><a name="push_back"></a>push_back Is Your Friend</div>
283<div class="doc_text">
284<p>
285One of the first things I noticed is the frequent use of the "push_back"
286method on the various lists. This is so common that it is worth mentioning.
287The "push_back" inserts a value into an STL list, vector, array, etc. at the
288end. The method might have also been named "insert_tail" or "append".
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]289Although I've used STL quite frequently, my use of push_back wasn't very
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]290high in other programs. In LLVM, you'll use it all the time.
291</p>
292</div>
293<!-- ======================================================================= -->
294<div class="doc_subsection"><a name="gep"></a>The Wily GetElementPtrInst</div>
295<div class="doc_text">
296<p>
297It took a little getting used to and several rounds of postings to the LLVM
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]298mailing list to wrap my head around this instruction correctly. Even though I had
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]299read the Language Reference and Programmer's Manual a couple times each, I still
300missed a few <em>very</em> key points:
301</p>
302<ul>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]303<li>GetElementPtrInst gives you back a Value for the last thing indexed.</li>
304<li>All global variables in LLVM are <em>pointers</em>.</li>
305<li>Pointers must also be dereferenced with the GetElementPtrInst
306instruction.</li>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]307</ul>
308<p>This means that when you look up an element in the global variable (assuming
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]309it's a struct or array), you <em>must</em> deference the pointer first! For many
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]310things, this leads to the idiom:
311</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]312<pre>
313std::vector&lt;Value*&gt; index_vector;
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]314index_vector.push_back( ConstantSInt::get( Type::LongTy, 0 );
315// ... push other indices ...
316GetElementPtrInst* gep = new GetElementPtrInst( ptr, index_vector );
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]317</pre>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]318<p>For example, suppose we have a global variable whose type is [24 x int]. The
319variable itself represents a <em>pointer</em> to that array. To subscript the
320array, we need two indices, not just one. The first index (0) dereferences the
321pointer. The second index subscripts the array. If you're a "C" programmer, this
322will run against your grain because you'll naturally think of the global array
323variable and the address of its first element as the same. That tripped me up
324for a while until I realized that they really do differ .. by <em>type</em>.
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]325Remember that LLVM is strongly typed. Everything has a type.
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]326The "type" of the global variable is [24 x int]*. That is, it's
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]327a pointer to an array of 24 ints. When you dereference that global variable with
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]328a single (0) index, you now have a "[24 x int]" type. Although
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]329the pointer value of the dereferenced global and the address of the zero'th element
330in the array will be the same, they differ in their type. The zero'th element has
331type "int" while the pointer value has type "[24 x int]".</p>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]332<p>Get this one aspect of LLVM right in your head, and you'll save yourself
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]333a lot of compiler writing headaches down the road.</p>
334</div>
335<!-- ======================================================================= -->
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]336<div class="doc_subsection"><a name="linkage"></a>Getting Linkage Types Right</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]337<div class="doc_text">
338<p>Linkage types in LLVM can be a little confusing, especially if your compiler
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]339writing mind has affixed firm concepts to particular words like "weak",
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]340"external", "global", "linkonce", etc. LLVM does <em>not</em> use the precise
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]341definitions of, say, ELF or GCC, even though they share common terms. To be fair,
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]342the concepts are related and similar but not precisely the same. This can lead
343you to think you know what a linkage type represents but in fact it is slightly
344different. I recommend you read the
345<a href="LangRef.html#linkage"> Language Reference on this topic</a> very
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]346carefully. Then, read it again.<p>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]347<p>Here are some handy tips that I discovered along the way:</p>
348<ul>
Alkis Evlogimenos0744b5f2004-03-11 10:14:21 +0000[diff] [blame]349 <li><em>Uninitialized means external.</em> That is, the symbol is declared in the current
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]350 module and can be used by that module, but it is not defined by that module.</li>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]351 <li><em>Setting an initializer changes a global' linkage type.</em> Setting an
352 initializer changes a global's linkage type from whatever it was to a normal,
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]353 defined global (not external). You'll need to call the setLinkage() method to
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]354 reset it if you specify the initializer after the GlobalValue has been constructed.
355 This is important for LinkOnce and Weak linkage types.</li>
356 <li><em>Appending linkage can keep track of things.</em> Appending linkage can
357 be used to keep track of compilation information at runtime. It could be used,
358 for example, to build a full table of all the C++ virtual tables or hold the
359 C++ RTTI data, or whatever. Appending linkage can only be applied to arrays.
360 All arrays with the same name in each module are concatenated together at link
361 time.</li>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]362</ul>
363</div>
364<!-- ======================================================================= -->
365<div class="doc_subsection"><a name="constants"></a>Constants Are Easier Than That!</div>
366<div class="doc_text">
367<p>
368Constants in LLVM took a little getting used to until I discovered a few utility
369functions in the LLVM IR that make things easier. Here's what I learned: </p>
370<ul>
371 <li>Constants are Values like anything else and can be operands of instructions</li>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]372 <li>Integer constants, frequently needed, can be created using the static "get"
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]373 methods of the ConstantInt, ConstantSInt, and ConstantUInt classes. The nice thing
374 about these is that you can "get" any kind of integer quickly.</li>
375 <li>There's a special method on Constant class which allows you to get the null
376 constant for <em>any</em> type. This is really handy for initializing large
377 arrays or structures, etc.</li>
378</ul>
379</div>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]380<!-- ======================================================================= -->
381<div class="doc_section"> <a name="lexicon">The Stacker Lexicon</a></div>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]382<div class="doc_text"><p>This section describes the Stacker language</p></div>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]383<div class="doc_subsection"><a name="stack"></a>The Stack</div>
384<div class="doc_text">
385<p>Stacker definitions define what they do to the global stack. Before
386proceeding, a few words about the stack are in order. The stack is simply
387a global array of 32-bit integers or pointers. A global index keeps track
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]388of the location of the top of the stack. All of this is hidden from the
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]389programmer, but it needs to be noted because it is the foundation of the
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]390conceptual programming model for Stacker. When you write a definition,
391you are, essentially, saying how you want that definition to manipulate
392the global stack.</p>
393<p>Manipulating the stack can be quite hazardous. There is no distinction
394given and no checking for the various types of values that can be placed
395on the stack. Automatic coercion between types is performed. In many
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]396cases, this is useful. For example, a boolean value placed on the stack
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]397can be interpreted as an integer with good results. However, using a
398word that interprets that boolean value as a pointer to a string to
399print out will almost always yield a crash. Stacker simply leaves it
400to the programmer to get it right without any interference or hindering
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]401on interpretation of the stack values. You've been warned. :) </p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]402</div>
403<!-- ======================================================================= -->
404<div class="doc_subsection"> <a name="punctuation"></a>Punctuation</div>
405<div class="doc_text">
406<p>Punctuation in Stacker is very simple. The colon and semi-colon
407characters are used to introduce and terminate a definition
408(respectively). Except for <em>FORWARD</em> declarations, definitions
409are all you can specify in Stacker. Definitions are read left to right.
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]410Immediately after the colon comes the name of the word being defined.
411The remaining words in the definition specify what the word does. The definition
412is terminated by a semi-colon.</p>
413<p>So, your typical definition will have the form:</p>
414<pre><code>: name ... ;</code></pre>
415<p>The <code>name</code> is up to you but it must start with a letter and contain
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]416only letters, numbers, and underscore. Names are case sensitive and must not be
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]417the same as the name of a built-in word. The <code>...</code> is replaced by
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]418the stack manipulating words that you wish to define <code>name</code> as. <p>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]419</div>
420<!-- ======================================================================= -->
421<div class="doc_subsection"><a name="comments"></a>Comments</div>
422<div class="doc_text">
423 <p>Stacker supports two types of comments. A hash mark (#) starts a comment
424 that extends to the end of the line. It is identical to the kind of comments
425 commonly used in shell scripts. A pair of parentheses also surround a comment.
426 In both cases, the content of the comment is ignored by the Stacker compiler. The
427 following does nothing in Stacker.
428 </p>
429<pre><code>
430# This is a comment to end of line
431( This is an enclosed comment )
432</code></pre>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]433<p>See the <a href="#example">example</a> program to see comments in use in
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]434a real program.</p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]435</div>
436<!-- ======================================================================= -->
437<div class="doc_subsection"><a name="literals"></a>Literals</div>
438<div class="doc_text">
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]439 <p>There are three kinds of literal values in Stacker: Integers, Strings,
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]440 and Booleans. In each case, the stack operation is to simply push the
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]441 value on to the stack. So, for example:<br/>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]442 <code> 42 " is the answer." TRUE </code><br/>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]443 will push three values on to the stack: the integer 42, the
444 string " is the answer.", and the boolean TRUE.</p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]445</div>
446<!-- ======================================================================= -->
447<div class="doc_subsection"><a name="words"></a>Words</div>
448<div class="doc_text">
449<p>Each definition in Stacker is composed of a set of words. Words are
450read and executed in order from left to right. There is very little
451checking in Stacker to make sure you're doing the right thing with
452the stack. It is assumed that the programmer knows how the stack
453transformation he applies will affect the program.</p>
454<p>Words in a definition come in two flavors: built-in and programmer
455defined. Simply mentioning the name of a previously defined or declared
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]456programmer-defined word causes that word's stack actions to be invoked. It
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]457is somewhat like a function call in other languages. The built-in
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]458words have various effects, described <a href="#builtins">below</a>.</p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]459<p>Sometimes you need to call a word before it is defined. For this, you can
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]460use the <code>FORWARD</code> declaration. It looks like this:</p>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]461<p><code>FORWARD name ;</code></p>
462<p>This simply states to Stacker that "name" is the name of a definition
463that is defined elsewhere. Generally it means the definition can be found
464"forward" in the file. But, it doesn't have to be in the current compilation
465unit. Anything declared with <code>FORWARD</code> is an external symbol for
466linking.</p>
467</div>
468<!-- ======================================================================= -->
469<div class="doc_subsection"><a name="builtins"></a>Built In Words</div>
470<div class="doc_text">
471<p>The built-in words of the Stacker language are put in several groups
472depending on what they do. The groups are as follows:</p>
473<ol>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]474 <li><em>Logical</em>: These words provide the logical operations for
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]475 comparing stack operands.<br/>The words are: &lt; &gt; &lt;= &gt;=
476 = &lt;&gt; true false.</li>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]477 <li><em>Bitwise</em>: These words perform bitwise computations on
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]478 their operands. <br/> The words are: &lt;&lt; &gt;&gt; XOR AND NOT</li>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]479 <li><em>Arithmetic</em>: These words perform arithmetic computations on
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]480 their operands. <br/> The words are: ABS NEG + - * / MOD */ ++ -- MIN MAX</li>
481 <li><em>Stack</em>These words manipulate the stack directly by moving
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]482 its elements around.<br/> The words are: DROP DROP2 NIP NIP2 DUP DUP2
483 SWAP SWAP2 OVER OVER2 ROT ROT2 RROT RROT2 TUCK TUCK2 PICK SELECT ROLL</li>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]484 <li><em>Memory</em>These words allocate, free, and manipulate memory
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]485 areas outside the stack.<br/>The words are: MALLOC FREE GET PUT</li>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]486 <li><em>Control</em>: These words alter the normal left to right flow
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]487 of execution.<br/>The words are: IF ELSE ENDIF WHILE END RETURN EXIT RECURSE</li>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]488 <li><em>I/O</em>: These words perform output on the standard output
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]489 and input on the standard input. No other I/O is possible in Stacker.
490 <br/>The words are: SPACE TAB CR &gt;s &gt;d &gt;c &lt;s &lt;d &lt;c.</li>
491</ol>
492<p>While you may be familiar with many of these operations from other
493programming languages, a careful review of their semantics is important
494for correct programming in Stacker. Of most importance is the effect
495that each of these built-in words has on the global stack. The effect is
496not always intuitive. To better describe the effects, we'll borrow from Forth the idiom of
497describing the effect on the stack with:</p>
498<p><code> BEFORE -- AFTER </code></p>
499<p>That is, to the left of the -- is a representation of the stack before
500the operation. To the right of the -- is a representation of the stack
501after the operation. In the table below that describes the operation of
502each of the built in words, we will denote the elements of the stack
503using the following construction:</p>
504<ol>
505 <li><em>b</em> - a boolean truth value</li>
506 <li><em>w</em> - a normal integer valued word.</li>
507 <li><em>s</em> - a pointer to a string value</li>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]508 <li><em>p</em> - a pointer to a malloc'd memory block</li>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]509</ol>
510</div>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]511<div class="doc_text" >
512 <table class="doc_table" style="border: 2px solid blue; border-collapse: collapse;" >
513<tr class="doc_table"><td colspan="4" style="border: 2px solid blue">Definition Of Operation Of Built In Words</td></tr>
514<tr class="doc_table"><td colspan="4" style="border: 2px solid blue"><b>LOGICAL OPERATIONS</b></td></tr>
515<tr class="doc_table">
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]516 <td style="border: 2px solid blue">Word</td>
517 <td style="border: 2px solid blue">Name</td>
518 <td style="border: 2px solid blue">Operation</td>
519 <td style="border: 2px solid blue">Description</td>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]520</tr>
521<tr class="doc_table"><td style="border: 2px solid blue">&lt;</td>
522 <td style="border: 2px solid blue">LT</td>
523 <td style="border: 2px solid blue">w1 w2 -- b</td>
524 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]525 compared. If w1 is less than w2, TRUE is pushed back on
526 the stack, otherwise FALSE is pushed back on the stack.</td>
527</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]528<tr><td style="border: 2px solid blue">&gt;</td>
529 <td style="border: 2px solid blue">GT</td>
530 <td style="border: 2px solid blue">w1 w2 -- b</td>
531 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]532 compared. If w1 is greater than w2, TRUE is pushed back on
533 the stack, otherwise FALSE is pushed back on the stack.</td>
534</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]535<tr><td style="border: 2px solid blue">&gt;=</td>
536 <td style="border: 2px solid blue">GE</td>
537 <td style="border: 2px solid blue">w1 w2 -- b</td>
538 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]539 compared. If w1 is greater than or equal to w2, TRUE is
540 pushed back on the stack, otherwise FALSE is pushed back
541 on the stack.</td>
542</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]543<tr><td style="border: 2px solid blue">&lt;=</td>
544 <td style="border: 2px solid blue">LE</td>
545 <td style="border: 2px solid blue">w1 w2 -- b</td>
546 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]547 compared. If w1 is less than or equal to w2, TRUE is
548 pushed back on the stack, otherwise FALSE is pushed back
549 on the stack.</td>
550</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]551<tr><td style="border: 2px solid blue">=</td>
552 <td style="border: 2px solid blue">EQ</td>
553 <td style="border: 2px solid blue">w1 w2 -- b</td>
554 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]555 compared. If w1 is equal to w2, TRUE is
556 pushed back on the stack, otherwise FALSE is pushed back
557 </td>
558</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]559<tr><td style="border: 2px solid blue">&lt;&gt;</td>
560 <td style="border: 2px solid blue">NE</td>
561 <td style="border: 2px solid blue">w1 w2 -- b</td>
562 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack and
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]563 compared. If w1 is equal to w2, TRUE is
564 pushed back on the stack, otherwise FALSE is pushed back
565 </td>
566</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]567<tr><td style="border: 2px solid blue">FALSE</td>
568 <td style="border: 2px solid blue">FALSE</td>
569 <td style="border: 2px solid blue"> -- b</td>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]570 <td style="border: 2px solid blue">The boolean value FALSE (0) is pushed on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]571</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]572<tr><td style="border: 2px solid blue">TRUE</td>
573 <td style="border: 2px solid blue">TRUE</td>
574 <td style="border: 2px solid blue"> -- b</td>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]575 <td style="border: 2px solid blue">The boolean value TRUE (-1) is pushed on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]576</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]577<tr><td colspan="4"><b>BITWISE OPERATORS</b></td></tr>
578<tr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]579 <td style="border: 2px solid blue">Word</td>
580 <td style="border: 2px solid blue">Name</td>
581 <td style="border: 2px solid blue">Operation</td>
582 <td style="border: 2px solid blue">Description</td>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]583</tr>
584<tr><td style="border: 2px solid blue">&lt;&lt;</td>
585 <td style="border: 2px solid blue">SHL</td>
586 <td style="border: 2px solid blue">w1 w2 -- w1&lt;&lt;w2</td>
587 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The w2
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]588 operand is shifted left by the number of bits given by the
589 w1 operand. The result is pushed back to the stack.</td>
590</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]591<tr><td style="border: 2px solid blue">&gt;&gt;</td>
592 <td style="border: 2px solid blue">SHR</td>
593 <td style="border: 2px solid blue">w1 w2 -- w1&gt;&gt;w2</td>
594 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The w2
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]595 operand is shifted right by the number of bits given by the
596 w1 operand. The result is pushed back to the stack.</td>
597</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]598<tr><td style="border: 2px solid blue">OR</td>
599 <td style="border: 2px solid blue">OR</td>
600 <td style="border: 2px solid blue">w1 w2 -- w2|w1</td>
601 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The values
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]602 are bitwise OR'd together and pushed back on the stack. This is
603 not a logical OR. The sequence 1 2 OR yields 3 not 1.</td>
604</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]605<tr><td style="border: 2px solid blue">AND</td>
606 <td style="border: 2px solid blue">AND</td>
607 <td style="border: 2px solid blue">w1 w2 -- w2&amp;w1</td>
608 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The values
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]609 are bitwise AND'd together and pushed back on the stack. This is
610 not a logical AND. The sequence 1 2 AND yields 0 not 1.</td>
611</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]612<tr><td style="border: 2px solid blue">XOR</td>
613 <td style="border: 2px solid blue">XOR</td>
614 <td style="border: 2px solid blue">w1 w2 -- w2^w1</td>
615 <td style="border: 2px solid blue">Two values (w1 and w2) are popped off the stack. The values
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]616 are bitwise exclusive OR'd together and pushed back on the stack.
617 For example, The sequence 1 3 XOR yields 2.</td>
618</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]619<tr><td colspan="4"><b>ARITHMETIC OPERATORS</b></td></tr>
620<tr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]621 <td style="border: 2px solid blue">Word</td>
622 <td style="border: 2px solid blue">Name</td>
623 <td style="border: 2px solid blue">Operation</td>
624 <td style="border: 2px solid blue">Description</td>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]625</tr>
626<tr><td style="border: 2px solid blue">ABS</td>
627 <td style="border: 2px solid blue">ABS</td>
628 <td style="border: 2px solid blue">w -- |w|</td>
629 <td style="border: 2px solid blue">One value s popped off the stack; its absolute value is computed
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]630 and then pushed on to the stack. If w1 is -1 then w2 is 1. If w1 is
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]631 1 then w2 is also 1.</td>
632</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]633<tr><td style="border: 2px solid blue">NEG</td>
634 <td style="border: 2px solid blue">NEG</td>
635 <td style="border: 2px solid blue">w -- -w</td>
636 <td style="border: 2px solid blue">One value is popped off the stack which is negated and then
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]637 pushed back on to the stack. If w1 is -1 then w2 is 1. If w1 is
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]638 1 then w2 is -1.</td>
639</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]640<tr><td style="border: 2px solid blue"> + </td>
641 <td style="border: 2px solid blue">ADD</td>
642 <td style="border: 2px solid blue">w1 w2 -- w2+w1</td>
643 <td style="border: 2px solid blue">Two values are popped off the stack. Their sum is pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]644 on to the stack</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]645</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]646<tr><td style="border: 2px solid blue"> - </td>
647 <td style="border: 2px solid blue">SUB</td>
648 <td style="border: 2px solid blue">w1 w2 -- w2-w1</td>
649 <td style="border: 2px solid blue">Two values are popped off the stack. Their difference is pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]650 on to the stack</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]651</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]652<tr><td style="border: 2px solid blue"> * </td>
653 <td style="border: 2px solid blue">MUL</td>
654 <td style="border: 2px solid blue">w1 w2 -- w2*w1</td>
655 <td style="border: 2px solid blue">Two values are popped off the stack. Their product is pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]656 on to the stack</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]657</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]658<tr><td style="border: 2px solid blue"> / </td>
659 <td style="border: 2px solid blue">DIV</td>
660 <td style="border: 2px solid blue">w1 w2 -- w2/w1</td>
661 <td style="border: 2px solid blue">Two values are popped off the stack. Their quotient is pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]662 on to the stack</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]663</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]664<tr><td style="border: 2px solid blue">MOD</td>
665 <td style="border: 2px solid blue">MOD</td>
666 <td style="border: 2px solid blue">w1 w2 -- w2%w1</td>
667 <td style="border: 2px solid blue">Two values are popped off the stack. Their remainder after division
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]668 of w1 by w2 is pushed back on to the stack</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]669</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]670<tr><td style="border: 2px solid blue"> */ </td>
671 <td style="border: 2px solid blue">STAR_SLAH</td>
672 <td style="border: 2px solid blue">w1 w2 w3 -- (w3*w2)/w1</td>
673 <td style="border: 2px solid blue">Three values are popped off the stack. The product of w1 and w2 is
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]674 divided by w3. The result is pushed back on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]675</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]676<tr><td style="border: 2px solid blue"> ++ </td>
677 <td style="border: 2px solid blue">INCR</td>
678 <td style="border: 2px solid blue">w -- w+1</td>
679 <td style="border: 2px solid blue">One value is popped off the stack. It is incremented by one and then
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]680 pushed back on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]681</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]682<tr><td style="border: 2px solid blue"> -- </td>
683 <td style="border: 2px solid blue">DECR</td>
684 <td style="border: 2px solid blue">w -- w-1</td>
685 <td style="border: 2px solid blue">One value is popped off the stack. It is decremented by one and then
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]686 pushed back on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]687</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]688<tr><td style="border: 2px solid blue">MIN</td>
689 <td style="border: 2px solid blue">MIN</td>
690 <td style="border: 2px solid blue">w1 w2 -- (w2&lt;w1?w2:w1)</td>
691 <td style="border: 2px solid blue">Two values are popped off the stack. The larger one is pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]692 on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]693</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]694<tr><td style="border: 2px solid blue">MAX</td>
695 <td style="border: 2px solid blue">MAX</td>
696 <td style="border: 2px solid blue">w1 w2 -- (w2&gt;w1?w2:w1)</td>
697 <td style="border: 2px solid blue">Two values are popped off the stack. The larger value is pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]698 on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]699</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]700<tr><td colspan="4"><b>STACK MANIPULATION OPERATORS</b></td></tr>
701<tr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]702 <td style="border: 2px solid blue">Word</td>
703 <td style="border: 2px solid blue">Name</td>
704 <td style="border: 2px solid blue">Operation</td>
705 <td style="border: 2px solid blue">Description</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]706</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]707<tr><td style="border: 2px solid blue">DROP</td>
708 <td style="border: 2px solid blue">DROP</td>
709 <td style="border: 2px solid blue">w -- </td>
710 <td style="border: 2px solid blue">One value is popped off the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]711</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]712<tr><td style="border: 2px solid blue">DROP2</td>
713 <td style="border: 2px solid blue">DROP2</td>
714 <td style="border: 2px solid blue">w1 w2 -- </td>
715 <td style="border: 2px solid blue">Two values are popped off the stack.</td>
716</tr>
717<tr><td style="border: 2px solid blue">NIP</td>
718 <td style="border: 2px solid blue">NIP</td>
719 <td style="border: 2px solid blue">w1 w2 -- w2</td>
720 <td style="border: 2px solid blue">The second value on the stack is removed from the stack. That is,
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]721 a value is popped off the stack and retained. Then a second value is
722 popped and the retained value is pushed.</td>
723</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]724<tr><td style="border: 2px solid blue">NIP2</td>
725 <td style="border: 2px solid blue">NIP2</td>
726 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w3 w4</td>
727 <td style="border: 2px solid blue">The third and fourth values on the stack are removed from it. That is,
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]728 two values are popped and retained. Then two more values are popped and
729 the two retained values are pushed back on.</td>
730</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]731<tr><td style="border: 2px solid blue">DUP</td>
732 <td style="border: 2px solid blue">DUP</td>
733 <td style="border: 2px solid blue">w1 -- w1 w1</td>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]734 <td style="border: 2px solid blue">One value is popped off the stack. That value is then pushed on to
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]735 the stack twice to duplicate the top stack vaue.</td>
736</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]737<tr><td style="border: 2px solid blue">DUP2</td>
738 <td style="border: 2px solid blue">DUP2</td>
739 <td style="border: 2px solid blue">w1 w2 -- w1 w2 w1 w2</td>
740 <td style="border: 2px solid blue">The top two values on the stack are duplicated. That is, two vaues
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]741 are popped off the stack. They are alternately pushed back on the
742 stack twice each.</td>
743</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]744<tr><td style="border: 2px solid blue">SWAP</td>
745 <td style="border: 2px solid blue">SWAP</td>
746 <td style="border: 2px solid blue">w1 w2 -- w2 w1</td>
747 <td style="border: 2px solid blue">The top two stack items are reversed in their order. That is, two
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]748 values are popped off the stack and pushed back on to the stack in
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]749 the opposite order they were popped.</td>
750</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]751<tr><td style="border: 2px solid blue">SWAP2</td>
752 <td style="border: 2px solid blue">SWAP2</td>
753 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w3 w4 w2 w1</td>
754 <td style="border: 2px solid blue">The top four stack items are swapped in pairs. That is, two values
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]755 are popped and retained. Then, two more values are popped and retained.
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]756 The values are pushed back on to the stack in the reverse order but
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]757 in pairs.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]758</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]759<tr><td style="border: 2px solid blue">OVER</td>
760 <td style="border: 2px solid blue">OVER</td>
761 <td style="border: 2px solid blue">w1 w2-- w1 w2 w1</td>
762 <td style="border: 2px solid blue">Two values are popped from the stack. They are pushed back
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]763 on to the stack in the order w1 w2 w1. This seems to cause the
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]764 top stack element to be duplicated "over" the next value.</td>
765</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]766<tr><td style="border: 2px solid blue">OVER2</td>
767 <td style="border: 2px solid blue">OVER2</td>
768 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w1 w2 w3 w4 w1 w2</td>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]769 <td style="border: 2px solid blue">The third and fourth values on the stack are replicated on to the
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]770 top of the stack</td>
771</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]772<tr><td style="border: 2px solid blue">ROT</td>
773 <td style="border: 2px solid blue">ROT</td>
774 <td style="border: 2px solid blue">w1 w2 w3 -- w2 w3 w1</td>
775 <td style="border: 2px solid blue">The top three values are rotated. That is, three value are popped
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]776 off the stack. They are pushed back on to the stack in the order
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]777 w1 w3 w2.</td>
778</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]779<tr><td style="border: 2px solid blue">ROT2</td>
780 <td style="border: 2px solid blue">ROT2</td>
781 <td style="border: 2px solid blue">w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2</td>
782 <td style="border: 2px solid blue">Like ROT but the rotation is done using three pairs instead of
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]783 three singles.</td>
784</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]785<tr><td style="border: 2px solid blue">RROT</td>
786 <td style="border: 2px solid blue">RROT</td>
787 <td style="border: 2px solid blue">w1 w2 w3 -- w2 w3 w1</td>
788 <td style="border: 2px solid blue">Reverse rotation. Like ROT, but it rotates the other way around.
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]789 Essentially, the third element on the stack is moved to the top
790 of the stack.</td>
791</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]792<tr><td style="border: 2px solid blue">RROT2</td>
793 <td style="border: 2px solid blue">RROT2</td>
794 <td style="border: 2px solid blue">w1 w2 w3 w4 w5 w6 -- w3 w4 w5 w6 w1 w2</td>
795 <td style="border: 2px solid blue">Double reverse rotation. Like RROT but the rotation is done using
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]796 three pairs instead of three singles. The fifth and sixth stack
797 elements are moved to the first and second positions</td>
798</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]799<tr><td style="border: 2px solid blue">TUCK</td>
800 <td style="border: 2px solid blue">TUCK</td>
801 <td style="border: 2px solid blue">w1 w2 -- w2 w1 w2</td>
802 <td style="border: 2px solid blue">Similar to OVER except that the second operand is being
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]803 replicated. Essentially, the first operand is being "tucked"
804 in between two instances of the second operand. Logically, two
805 values are popped off the stack. They are placed back on the
806 stack in the order w2 w1 w2.</td>
807</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]808<tr><td style="border: 2px solid blue">TUCK2</td>
809 <td style="border: 2px solid blue">TUCK2</td>
810 <td style="border: 2px solid blue">w1 w2 w3 w4 -- w3 w4 w1 w2 w3 w4</td>
811 <td style="border: 2px solid blue">Like TUCK but a pair of elements is tucked over two pairs.
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]812 That is, the top two elements of the stack are duplicated and
813 inserted into the stack at the fifth and positions.</td>
814</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]815<tr><td style="border: 2px solid blue">PICK</td>
816 <td style="border: 2px solid blue">PICK</td>
817 <td style="border: 2px solid blue">x0 ... Xn n -- x0 ... Xn x0</td>
818 <td style="border: 2px solid blue">The top of the stack is used as an index into the remainder of
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]819 the stack. The element at the nth position replaces the index
820 (top of stack). This is useful for cycling through a set of
821 values. Note that indexing is zero based. So, if n=0 then you
822 get the second item on the stack. If n=1 you get the third, etc.
823 Note also that the index is replaced by the n'th value. </td>
824</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]825<tr><td style="border: 2px solid blue">SELECT</td>
826 <td style="border: 2px solid blue">SELECT</td>
827 <td style="border: 2px solid blue">m n X0..Xm Xm+1 .. Xn -- Xm</td>
828 <td style="border: 2px solid blue">This is like PICK but the list is removed and you need to specify
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]829 both the index and the size of the list. Careful with this one,
830 the wrong value for n can blow away a huge amount of the stack.</td>
831</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]832<tr><td style="border: 2px solid blue">ROLL</td>
833 <td style="border: 2px solid blue">ROLL</td>
834 <td style="border: 2px solid blue">x0 x1 .. xn n -- x1 .. xn x0</td>
835 <td style="border: 2px solid blue"><b>Not Implemented</b>. This one has been left as an exercise to
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]836 the student. See <a href="#exercise">Exercise</a>. ROLL requires
837 a value, "n", to be on the top of the stack. This value specifies how
838 far into the stack to "roll". The n'th value is <em>moved</em> (not
839 copied) from its location and replaces the "n" value on the top of the
840 stack. In this way, all the values between "n" and x0 roll up the stack.
841 The operation of ROLL is a generalized ROT. The "n" value specifies
842 how much to rotate. That is, ROLL with n=1 is the same as ROT and
843 ROLL with n=2 is the same as ROT2.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]844</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]845<tr><td colspan="4"><b>MEMORY OPERATORS</b></td></tr>
846<tr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]847 <td style="border: 2px solid blue">Word</td>
848 <td style="border: 2px solid blue">Name</td>
849 <td style="border: 2px solid blue">Operation</td>
850 <td style="border: 2px solid blue">Description</td>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]851</tr>
852<tr><td style="border: 2px solid blue">MALLOC</td>
853 <td style="border: 2px solid blue">MALLOC</td>
854 <td style="border: 2px solid blue">w1 -- p</td>
855 <td style="border: 2px solid blue">One value is popped off the stack. The value is used as the size
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]856 of a memory block to allocate. The size is in bytes, not words.
857 The memory allocation is completed and the address of the memory
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]858 block is pushed on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]859</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]860<tr><td style="border: 2px solid blue">FREE</td>
861 <td style="border: 2px solid blue">FREE</td>
862 <td style="border: 2px solid blue">p -- </td>
863 <td style="border: 2px solid blue">One pointer value is popped off the stack. The value should be
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]864 the address of a memory block created by the MALLOC operation. The
865 associated memory block is freed. Nothing is pushed back on the
866 stack. Many bugs can be created by attempting to FREE something
867 that isn't a pointer to a MALLOC allocated memory block. Make
868 sure you know what's on the stack. One way to do this is with
869 the following idiom:<br/>
870 <code>64 MALLOC DUP DUP (use ptr) DUP (use ptr) ... FREE</code>
871 <br/>This ensures that an extra copy of the pointer is placed on
872 the stack (for the FREE at the end) and that every use of the
873 pointer is preceded by a DUP to retain the copy for FREE.</td>
874</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]875<tr><td style="border: 2px solid blue">GET</td>
876 <td style="border: 2px solid blue">GET</td>
877 <td style="border: 2px solid blue">w1 p -- w2 p</td>
878 <td style="border: 2px solid blue">An integer index and a pointer to a memory block are popped of
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]879 the block. The index is used to index one byte from the memory
880 block. That byte value is retained, the pointer is pushed again
881 and the retained value is pushed. Note that the pointer value
882 s essentially retained in its position so this doesn't count
883 as a "use ptr" in the FREE idiom.</td>
884</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]885<tr><td style="border: 2px solid blue">PUT</td>
886 <td style="border: 2px solid blue">PUT</td>
887 <td style="border: 2px solid blue">w1 w2 p -- p </td>
888 <td style="border: 2px solid blue">An integer value is popped of the stack. This is the value to
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]889 be put into a memory block. Another integer value is popped of
890 the stack. This is the indexed byte in the memory block. A
891 pointer to the memory block is popped off the stack. The
892 first value (w1) is then converted to a byte and written
893 to the element of the memory block(p) at the index given
894 by the second value (w2). The pointer to the memory block is
895 pushed back on the stack so this doesn't count as a "use ptr"
896 in the FREE idiom.</td>
897</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]898<tr><td colspan="4"><b>CONTROL FLOW OPERATORS</b></td></tr>
899<tr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]900 <td style="border: 2px solid blue">Word</td>
901 <td style="border: 2px solid blue">Name</td>
902 <td style="border: 2px solid blue">Operation</td>
903 <td style="border: 2px solid blue">Description</td>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]904</tr>
905<tr><td style="border: 2px solid blue">RETURN</td>
906 <td style="border: 2px solid blue">RETURN</td>
907 <td style="border: 2px solid blue"> -- </td>
908 <td style="border: 2px solid blue">The currently executing definition returns immediately to its caller.
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]909 Note that there is an implicit <code>RETURN</code> at the end of each
910 definition, logically located at the semi-colon. The sequence
911 <code>RETURN ;</code> is valid but redundant.</td>
912</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]913<tr><td style="border: 2px solid blue">EXIT</td>
914 <td style="border: 2px solid blue">EXIT</td>
915 <td style="border: 2px solid blue">w1 -- </td>
916 <td style="border: 2px solid blue">A return value for the program is popped off the stack. The program is
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]917 then immediately terminated. This is normally an abnormal exit from the
918 program. For a normal exit (when <code>MAIN</code> finishes), the exit
919 code will always be zero in accordance with UNIX conventions.</td>
920</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]921<tr><td style="border: 2px solid blue">RECURSE</td>
922 <td style="border: 2px solid blue">RECURSE</td>
923 <td style="border: 2px solid blue"> -- </td>
924 <td style="border: 2px solid blue">The currently executed definition is called again. This operation is
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]925 needed since the definition of a word doesn't exist until the semi colon
926 is reacher. Attempting something like:<br/>
927 <code> : recurser recurser ; </code><br/> will yield and error saying that
928 "recurser" is not defined yet. To accomplish the same thing, change this
929 to:<br/>
930 <code> : recurser RECURSE ; </code></td>
931</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]932<tr><td style="border: 2px solid blue">IF (words...) ENDIF</td>
933 <td style="border: 2px solid blue">IF (words...) ENDIF</td>
934 <td style="border: 2px solid blue">b -- </td>
935 <td style="border: 2px solid blue">A boolean value is popped of the stack. If it is non-zero then the "words..."
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]936 are executed. Otherwise, execution continues immediately following the ENDIF.</td>
937</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]938<tr><td style="border: 2px solid blue">IF (words...) ELSE (words...) ENDIF</td>
939 <td style="border: 2px solid blue">IF (words...) ELSE (words...) ENDIF</td>
940 <td style="border: 2px solid blue">b -- </td>
941 <td style="border: 2px solid blue">A boolean value is popped of the stack. If it is non-zero then the "words..."
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]942 between IF and ELSE are executed. Otherwise the words between ELSE and ENDIF are
943 executed. In either case, after the (words....) have executed, execution continues
944 immediately following the ENDIF. </td>
945</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]946<tr><td style="border: 2px solid blue">WHILE (words...) END</td>
947 <td style="border: 2px solid blue">WHILE (words...) END</td>
948 <td style="border: 2px solid blue">b -- b </td>
949 <td style="border: 2px solid blue">The boolean value on the top of the stack is examined. If it is non-zero then the
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]950 "words..." between WHILE and END are executed. Execution then begins again at the WHILE where another
951 boolean is popped off the stack. To prevent this operation from eating up the entire
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]952 stack, you should push on to the stack (just before the END) a boolean value that indicates
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]953 whether to terminate. Note that since booleans and integers can be coerced you can
954 use the following "for loop" idiom:<br/>
955 <code>(push count) WHILE (words...) -- END</code><br/>
956 For example:<br/>
957 <code>10 WHILE DUP &gt;d -- END</code><br/>
958 This will print the numbers from 10 down to 1. 10 is pushed on the stack. Since that is
959 non-zero, the while loop is entered. The top of the stack (10) is duplicated and then
960 printed out with &gt;d. The top of the stack is decremented, yielding 9 and control is
961 transfered back to the WHILE keyword. The process starts all over again and repeats until
962 the top of stack is decremented to 0 at which the WHILE test fails and control is
963 transfered to the word after the END.</td>
964</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]965<tr><td colspan="4"><b>INPUT &amp; OUTPUT OPERATORS</b></td></tr>
966<tr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]967 <td style="border: 2px solid blue">Word</td>
968 <td style="border: 2px solid blue">Name</td>
969 <td style="border: 2px solid blue">Operation</td>
970 <td style="border: 2px solid blue">Description</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]971</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]972<tr><td style="border: 2px solid blue">SPACE</td>
973 <td style="border: 2px solid blue">SPACE</td>
974 <td style="border: 2px solid blue"> -- </td>
975 <td style="border: 2px solid blue">A space character is put out. There is no stack effect.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]976</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]977<tr><td style="border: 2px solid blue">TAB</td>
978 <td style="border: 2px solid blue">TAB</td>
979 <td style="border: 2px solid blue"> -- </td>
980 <td style="border: 2px solid blue">A tab character is put out. There is no stack effect.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]981</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]982<tr><td style="border: 2px solid blue">CR</td>
983 <td style="border: 2px solid blue">CR</td>
984 <td style="border: 2px solid blue"> -- </td>
985 <td style="border: 2px solid blue">A carriage return character is put out. There is no stack effect.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]986</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]987<tr><td style="border: 2px solid blue">&gt;s</td>
988 <td style="border: 2px solid blue">OUT_STR</td>
989 <td style="border: 2px solid blue"> -- </td>
990 <td style="border: 2px solid blue">A string pointer is popped from the stack. It is put out.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]991</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]992<tr><td style="border: 2px solid blue">&gt;d</td>
993 <td style="border: 2px solid blue">OUT_STR</td>
994 <td style="border: 2px solid blue"> -- </td>
995 <td style="border: 2px solid blue">A value is popped from the stack. It is put out as a decimal integer.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]996</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]997<tr><td style="border: 2px solid blue">&gt;c</td>
998 <td style="border: 2px solid blue">OUT_CHR</td>
999 <td style="border: 2px solid blue"> -- </td>
1000 <td style="border: 2px solid blue">A value is popped from the stack. It is put out as an ASCII character.</td>
1001</tr>
1002<tr><td style="border: 2px solid blue">&lt;s</td>
1003 <td style="border: 2px solid blue">IN_STR</td>
1004 <td style="border: 2px solid blue"> -- s </td>
1005 <td style="border: 2px solid blue">A string is read from the input via the scanf(3) format string " %as". The
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1006 resulting string is pushed on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1007</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1008<tr><td style="border: 2px solid blue">&lt;d</td>
1009 <td style="border: 2px solid blue">IN_STR</td>
1010 <td style="border: 2px solid blue"> -- w </td>
1011 <td style="border: 2px solid blue">An integer is read from the input via the scanf(3) format string " %d". The
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1012 resulting value is pushed on to the stack</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1013</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1014<tr><td style="border: 2px solid blue">&lt;c</td>
1015 <td style="border: 2px solid blue">IN_CHR</td>
1016 <td style="border: 2px solid blue"> -- w </td>
1017 <td style="border: 2px solid blue">A single character is read from the input via the scanf(3) format string
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1018 " %c". The value is converted to an integer and pushed on to the stack.</td>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1019</tr>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1020<tr><td style="border: 2px solid blue">DUMP</td>
1021 <td style="border: 2px solid blue">DUMP</td>
1022 <td style="border: 2px solid blue"> -- </td>
1023 <td style="border: 2px solid blue">The stack contents are dumped to standard output. This is useful for
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1024 debugging your definitions. Put DUMP at the beginning and end of a definition
1025 to see instantly the net effect of the definition.</td>
1026</tr>
1027</table>
1028</div>
1029<!-- ======================================================================= -->
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1030<div class="doc_section"> <a name="example">Prime: A Complete Example</a></div>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1031<div class="doc_text">
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1032<p>The following fully documented program highlights many features of both
1033the Stacker language and what is possible with LLVM. The program has two modes
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1034of operation. If you provide numeric arguments to the program, it checks to see
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1035if those arguments are prime numbers and prints out the results. Without any
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1036arguments, the program prints out any prime numbers it finds between 1 and one
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1037million (there's a lot of them!). The source code comments below tell the
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1038remainder of the story.
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1039</p>
1040</div>
1041<div class="doc_text">
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1042<pre><code>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1043################################################################################
1044#
1045# Brute force prime number generator
1046#
1047# This program is written in classic Stacker style, that being the style of a
1048# stack. Start at the bottom and read your way up !
1049#
1050# Reid Spencer - Nov 2003
1051################################################################################
1052# Utility definitions
1053################################################################################
1054: print >d CR ;
1055: it_is_a_prime TRUE ;
1056: it_is_not_a_prime FALSE ;
1057: continue_loop TRUE ;
1058: exit_loop FALSE;
1059
1060################################################################################
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1061# This definition tries an actual division of a candidate prime number. It
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1062# determines whether the division loop on this candidate should continue or
1063# not.
1064# STACK<:
1065# div - the divisor to try
1066# p - the prime number we are working on
1067# STACK>:
1068# cont - should we continue the loop ?
1069# div - the next divisor to try
1070# p - the prime number we are working on
1071################################################################################
1072: try_dividing
1073 DUP2 ( save div and p )
1074 SWAP ( swap to put divisor second on stack)
1075 MOD 0 = ( get remainder after division and test for 0 )
1076 IF
1077 exit_loop ( remainder = 0, time to exit )
1078 ELSE
1079 continue_loop ( remainder != 0, keep going )
1080 ENDIF
1081;
1082
1083################################################################################
1084# This function tries one divisor by calling try_dividing. But, before doing
1085# that it checks to see if the value is 1. If it is, it does not bother with
1086# the division because prime numbers are allowed to be divided by one. The
1087# top stack value (cont) is set to determine if the loop should continue on
1088# this prime number or not.
1089# STACK<:
1090# cont - should we continue the loop (ignored)?
1091# div - the divisor to try
1092# p - the prime number we are working on
1093# STACK>:
1094# cont - should we continue the loop ?
1095# div - the next divisor to try
1096# p - the prime number we are working on
1097################################################################################
1098: try_one_divisor
1099 DROP ( drop the loop continuation )
1100 DUP ( save the divisor )
1101 1 = IF ( see if divisor is == 1 )
1102 exit_loop ( no point dividing by 1 )
1103 ELSE
1104 try_dividing ( have to keep going )
1105 ENDIF
1106 SWAP ( get divisor on top )
1107 -- ( decrement it )
1108 SWAP ( put loop continuation back on top )
1109;
1110
1111################################################################################
1112# The number on the stack (p) is a candidate prime number that we must test to
1113# determine if it really is a prime number. To do this, we divide it by every
1114# number from one p-1 to 1. The division is handled in the try_one_divisor
1115# definition which returns a loop continuation value (which we also seed with
1116# the value 1). After the loop, we check the divisor. If it decremented all
1117# the way to zero then we found a prime, otherwise we did not find one.
1118# STACK<:
1119# p - the prime number to check
1120# STACK>:
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1121# yn - boolean indicating if its a prime or not
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1122# p - the prime number checked
1123################################################################################
1124: try_harder
1125 DUP ( duplicate to get divisor value ) )
1126 -- ( first divisor is one less than p )
1127 1 ( continue the loop )
1128 WHILE
1129 try_one_divisor ( see if its prime )
1130 END
1131 DROP ( drop the continuation value )
1132 0 = IF ( test for divisor == 1 )
1133 it_is_a_prime ( we found one )
1134 ELSE
1135 it_is_not_a_prime ( nope, this one is not a prime )
1136 ENDIF
1137;
1138
1139################################################################################
1140# This definition determines if the number on the top of the stack is a prime
1141# or not. It does this by testing if the value is degenerate (<= 3) and
1142# responding with yes, its a prime. Otherwise, it calls try_harder to actually
1143# make some calculations to determine its primeness.
1144# STACK<:
1145# p - the prime number to check
1146# STACK>:
1147# yn - boolean indicating if its a prime or not
1148# p - the prime number checked
1149################################################################################
1150: is_prime
1151 DUP ( save the prime number )
1152 3 >= IF ( see if its <= 3 )
1153 it_is_a_prime ( its <= 3 just indicate its prime )
1154 ELSE
1155 try_harder ( have to do a little more work )
1156 ENDIF
1157;
1158
1159################################################################################
1160# This definition is called when it is time to exit the program, after we have
1161# found a sufficiently large number of primes.
1162# STACK<: ignored
1163# STACK>: exits
1164################################################################################
1165: done
1166 "Finished" >s CR ( say we are finished )
1167 0 EXIT ( exit nicely )
1168;
1169
1170################################################################################
1171# This definition checks to see if the candidate is greater than the limit. If
1172# it is, it terminates the program by calling done. Otherwise, it increments
1173# the value and calls is_prime to determine if the candidate is a prime or not.
1174# If it is a prime, it prints it. Note that the boolean result from is_prime is
1175# gobbled by the following IF which returns the stack to just contining the
1176# prime number just considered.
1177# STACK<:
1178# p - one less than the prime number to consider
1179# STACK>
1180# p+1 - the prime number considered
1181################################################################################
1182: consider_prime
1183 DUP ( save the prime number to consider )
1184 1000000 < IF ( check to see if we are done yet )
1185 done ( we are done, call "done" )
1186 ENDIF
1187 ++ ( increment to next prime number )
1188 is_prime ( see if it is a prime )
1189 IF
1190 print ( it is, print it )
1191 ENDIF
1192;
1193
1194################################################################################
1195# This definition starts at one, prints it out and continues into a loop calling
1196# consider_prime on each iteration. The prime number candidate we are looking at
1197# is incremented by consider_prime.
1198# STACK<: empty
1199# STACK>: empty
1200################################################################################
1201: find_primes
1202 "Prime Numbers: " >s CR ( say hello )
1203 DROP ( get rid of that pesky string )
1204 1 ( stoke the fires )
1205 print ( print the first one, we know its prime )
1206 WHILE ( loop while the prime to consider is non zero )
1207 consider_prime ( consider one prime number )
1208 END
1209;
1210
1211################################################################################
1212#
1213################################################################################
1214: say_yes
1215 >d ( Print the prime number )
1216 " is prime." ( push string to output )
1217 >s ( output it )
1218 CR ( print carriage return )
1219 DROP ( pop string )
1220;
1221
1222: say_no
1223 >d ( Print the prime number )
1224 " is NOT prime." ( push string to put out )
1225 >s ( put out the string )
1226 CR ( print carriage return )
1227 DROP ( pop string )
1228;
1229
1230################################################################################
1231# This definition processes a single command line argument and determines if it
1232# is a prime number or not.
1233# STACK<:
1234# n - number of arguments
1235# arg1 - the prime numbers to examine
1236# STACK>:
1237# n-1 - one less than number of arguments
1238# arg2 - we processed one argument
1239################################################################################
1240: do_one_argument
1241 -- ( decrement loop counter )
1242 SWAP ( get the argument value )
1243 is_prime IF ( determine if its prime )
1244 say_yes ( uhuh )
1245 ELSE
1246 say_no ( nope )
1247 ENDIF
1248 DROP ( done with that argument )
1249;
1250
1251################################################################################
1252# The MAIN program just prints a banner and processes its arguments.
1253# STACK<:
1254# n - number of arguments
1255# ... - the arguments
1256################################################################################
1257: process_arguments
1258 WHILE ( while there are more arguments )
1259 do_one_argument ( process one argument )
1260 END
1261;
1262
1263################################################################################
1264# The MAIN program just prints a banner and processes its arguments.
1265# STACK<: arguments
1266################################################################################
1267: MAIN
1268 NIP ( get rid of the program name )
1269 -- ( reduce number of arguments )
1270 DUP ( save the arg counter )
1271 1 <= IF ( See if we got an argument )
1272 process_arguments ( tell user if they are prime )
1273 ELSE
1274 find_primes ( see how many we can find )
1275 ENDIF
1276 0 ( push return code )
1277;
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1278</code>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1279</pre>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1280</div>
1281<!-- ======================================================================= -->
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1282<div class="doc_section"> <a name="internal">Internals</a></div>
1283<div class="doc_text">
1284 <p><b>This section is under construction.</b>
1285 <p>In the mean time, you can always read the code! It has comments!</p>
1286</div>
1287<!-- ======================================================================= -->
1288<div class="doc_subsection"> <a name="directory">Directory Structure</a></div>
1289<div class="doc_text">
1290<p>The source code, test programs, and sample programs can all be found
1291under the LLVM "projects" directory. You will need to obtain the LLVM sources
1292to find it (either via anonymous CVS or a tarball. See the
1293<a href="GettingStarted.html">Getting Started</a> document).</p>
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1294<p>Under the "projects" directory there is a directory named "Stacker". That
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1295directory contains everything, as follows:</p>
1296<ul>
1297 <li><em>lib</em> - contains most of the source code
1298 <ul>
1299 <li><em>lib/compiler</em> - contains the compiler library
1300 <li><em>lib/runtime</em> - contains the runtime library
1301 </ul></li>
1302 <li><em>test</em> - contains the test programs</li>
1303 <li><em>tools</em> - contains the Stacker compiler main program, stkrc
1304 <ul>
1305 <li><em>lib/stkrc</em> - contains the Stacker compiler main program
1306 </ul</li>
1307 <li><em>sample</em> - contains the sample programs</li>
1308</ul>
1309</div>
1310<!-- ======================================================================= -->
1311<div class="doc_subsection"><a name="lexer"></a>The Lexer</div>
1312<div class="doc_text">
1313<p>See projects/Stacker/lib/compiler/Lexer.l</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1314</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1315<!-- ======================================================================= -->
1316<div class="doc_subsection"><a name="parser"></a>The Parser</div>
1317<div class="doc_text">
1318<p>See projects/Stacker/lib/compiler/StackerParser.y</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1319</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1320<!-- ======================================================================= -->
1321<div class="doc_subsection"><a name="compiler"></a>The Compiler</div>
1322<div class="doc_text">
1323<p>See projects/Stacker/lib/compiler/StackerCompiler.cpp</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1324</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1325<!-- ======================================================================= -->
1326<div class="doc_subsection"><a name="runtime"></a>The Runtime</div>
1327<div class="doc_text">
1328<p>See projects/Stacker/lib/runtime/stacker_rt.c</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1329</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1330<!-- ======================================================================= -->
1331<div class="doc_subsection"><a name="driver"></a>Compiler Driver</div>
1332<div class="doc_text">
1333<p>See projects/Stacker/tools/stkrc/stkrc.cpp</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1334</div>
Brian Gaeke07e89e42003-11-24 17:03:38 +0000[diff] [blame]1335<!-- ======================================================================= -->
1336<div class="doc_subsection"><a name="tests"></a>Test Programs</div>
1337<div class="doc_text">
1338<p>See projects/Stacker/test/*.st</p>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1339</div>
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1340<!-- ======================================================================= -->
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1341<div class="doc_subsection"> <a name="exercise">Exercise</a></div>
1342<div class="doc_text">
1343<p>As you may have noted from a careful inspection of the Built-In word
1344definitions, the ROLL word is not implemented. This word was left out of
1345Stacker on purpose so that it can be an exercise for the student. The exercise
1346is to implement the ROLL functionality (in your own workspace) and build a test
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1347program for it. If you can implement ROLL, you understand Stacker and probably
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1348a fair amount about LLVM since this is one of the more complicated Stacker
1349operations. The work will almost be completely limited to the
1350<a href="#compiler">compiler</a>.
1351<p>The ROLL word is already recognized by both the lexer and parser but ignored
1352by the compiler. That means you don't have to futz around with figuring out how
1353to get the keyword recognized. It already is. The part of the compiler that
1354you need to implement is the <code>ROLL</code> case in the
Misha Brukmanfe22af62004-04-16 16:20:07 +0000[diff] [blame]1355<code>StackerCompiler::handle_word(int)</code> method.</p> See the
1356implementations of PICK and SELECT in the same method to get some hints about
1357how to complete this exercise.<p>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1358<p>Good luck!</p>
1359</div>
1360<!-- ======================================================================= -->
Misha Brukmanfe22af62004-04-16 16:20:07 +0000[diff] [blame]1361<div class="doc_subsection"><a name="todo">Things Remaining To Be Done</a></div>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1362<div class="doc_text">
1363<p>The initial implementation of Stacker has several deficiencies. If you're
1364interested, here are some things that could be implemented better:</p>
1365<ol>
1366 <li>Write an LLVM pass to compute the correct stack depth needed by the
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1367 program. Currently the stack is set to a fixed number which means programs
1368 with large numbers of definitions might fail.</li>
1369 <li>Enhance to run on 64-bit platforms like SPARC. Right now the size of a
Misha Brukmanfe22af62004-04-16 16:20:07 +0000[diff] [blame]1370 pointer on 64-bit machines will cause incorrect results because of the
1371 32-bit size of a stack element currently supported. This feature was not
1372 implemented because LLVM needs a union type to be able to support the
1373 different sizes correctly (portably and efficiently).</li>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1374 <li>Write an LLVM pass to optimize the use of the global stack. The code
1375 emitted currently is somewhat wasteful. It gets cleaned up a lot by existing
1376 passes but more could be done.</li>
1377 <li>Add -O -O1 -O2 and -O3 optimization switches to the compiler driver to
John Criswelld000e1d2003-12-18 16:43:17 +0000[diff] [blame]1378 allow LLVM optimization without using "opt."</li>
Misha Brukmanfe22af62004-04-16 16:20:07 +0000[diff] [blame]1379 <li>Make the compiler driver use the LLVM linking facilities (with IPO)
1380 before depending on GCC to do the final link.</li>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1381 <li>Clean up parsing. It doesn't handle errors very well.</li>
1382 <li>Rearrange the StackerCompiler.cpp code to make better use of inserting
1383 instructions before a block's terminating instruction. I didn't figure this
Misha Brukmanfe22af62004-04-16 16:20:07 +0000[diff] [blame]1384 technique out until I was nearly done with LLVM. As it is, its a bad example
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1385 of how to insert instructions!</li>
1386 <li>Provide for I/O to arbitrary files instead of just stdin/stdout.</li>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1387 <li>Write additional built-in words; with inspiration from FORTH</li>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1388 <li>Write additional sample Stacker programs.</li>
Chris Lattner45ab10c2003-12-18 06:40:22 +0000[diff] [blame]1389 <li>Add your own compiler writing experiences and tips in the
1390 <a href="#lessons">Lessons I Learned About LLVM</a> section.</li>
Chris Lattnere46d6012003-11-25 01:35:06 +0000[diff] [blame]1391</ol>
1392</div>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1393
1394<!-- *********************************************************************** -->
1395
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1396<hr>
Misha Brukman36692992004-05-12 19:52:00 +0000[diff] [blame^]1397<address>
1398 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1399 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
1400 <a href="http://validator.w3.org/check/referer"><img
1401 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
1402
1403 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a><br>
1404 <a href="http://llvm.cs.uiuc.edu">LLVM Compiler Infrastructure</a><br>
1405 Last modified: $Date$
1406</address>
1407
Brian Gaeke90181482003-11-24 02:52:51 +0000[diff] [blame]1408</body>
1409</html>